How to set a useful stop rule for a marketing experiment

Understand the decision

Define the question, planned observation window and decision measure with the budget owner. Identify reasons the result could be uninterpretable, such as missing consent data or an unavailable form. Avoid changing the rules repeatedly to justify a preferred outcome. A test that reveals insufficient evidence can still be useful if that conclusion is recorded honestly.

A practical approach

Define the audience, offer, action and follow-up before choosing a channel. Make the promise in the campaign consistent with what the landing page and receiving team can deliver. Separate creative testing from audience or offer changes so the result remains interpretable. Work with an agreed budget and stopping rule. Use the business’s own evidence to judge fit; avoid borrowing benchmark conversion rates from a different market.

• Write continue, pause and change conditions.
• Include journey failures and capacity limits.
• Record the observation scope.
• Review uncertainty before extending spend.

Illustrative example

A consultancy testing a new enquiry offer could pause if the destination fails or requests repeatedly reach the wrong team. At review, it would assess fit and the quality of the next conversation. It would not automatically scale because a small number of clicks looked promising.

A mistake to avoid

Do not let the absence of a clear stopping rule turn an experiment into indefinite spending. Learning requires a decision, even when the evidence remains limited.

What to review

Track whether the experiment answered its question, respected the budget and produced a documented next decision rather than only an activity report.

Turn the guide into a working brief

Keep a campaign brief with the offer, exclusions, approved claims, destination, measurement plan and named response owner. Test the destination and receipt before sending paid traffic. Review enquiry quality with the receiving team alongside clicks and spend. Document what changed and why between campaign versions. Scale only when the delivery and follow-up process can support additional demand without losing the information customers need.

Common questions